Institutional variability in transfusion practice for liver transplantation.

We prospectively evaluated the institutional variability in perioperative transfusion therapy in orthotopic liver transplantation (OLT). Adult OLTs completed during a 12-mo period were studied until the 48th postoperative hour at 8 centers. A multivariate analysis using mixed-effects logistic regression included variables predisposing to blood loss and a center random effect. In addition, the influence of the calculated perioperative hemoglobin (Hb) loss on the individual probability of receiving red blood cells (RBCs), fresh frozen plasma (FFP), and platelets in excess of the overall median were explored. The analysis was performed on 301 cases. The overall median numbers transfused were 5 RBC units, 6 FFP units, and the median platelet dose was 5.10(11), with significant intercentric differences in the proportions of cases given more than the overall median. Intercentric differences remained significant after adjustment for factors independently associated with a large blood component use. Intercentric differences in RBCs, FFP, and platelet use decreased but persisted after adjustment for the perioperative Hb loss. Intercentric differences in RBC use disappeared after adjustment for the postoperative Hb concentration. The significant heterogeneity in transfusion therapy mandates reassessment of the rational use of blood products in OLT.